Java vs Golang - Parameterized tests

Although I like Java, I think that with all its drawback it brings years of knowledge and optimization, I find too many times that when you "think in Java" you tend to creat complicated solutions. By complicated I mean hard for a developer to grasp and to remember.

In the following example we can see several differences that make the Java code look `old` and clumsy and not fun.
1. 7 imports  VS 1 in Golang.
2. Everything is a class. Fibonacci is a function in Golang.
3. Why hiding from the developer the fact that the test runs in a loop?
4. All these annotations.

Simplicity matter!
When you try to produce a simple code you end up with less dependency, less code, less magic, less moving parts and less things to know in order to produce quality code.
And as always, but especially in programming, Less Is More.


Simplicity Matters by Rich Hickey

The Java way from JUnit-team

 import static org.junit.Assert.assertEquals;  
 import java.util.Arrays;  
 import java.util.Collection;  
 import org.junit.Test;  
 import org.junit.runner.RunWith;  
 import org.junit.runners.Parameterized;  
 import org.junit.runners.Parameterized.Parameters;  
 @RunWith(Parameterized.class)  
 public class FibonacciTest {  
   @Parameters  
   public static Collection<Object[]> data() {  
     return Arrays.asList(new Object[][] {     
          { 0, 0 }, { 1, 1 }, { 2, 1 }, { 3, 2 }, { 4, 3 }, { 5, 5 }, { 6, 8 }   
       });  
   }  
   private int fInput;  
   private int fExpected;  
   public FibonacciTest(int input, int expected) {  
     fInput= input;  
     fExpected= expected;  
   }  
   @Test  
   public void test() {  
     assertEquals(fExpected, Fibonacci.compute(fInput));  
   }  
 }  
 public class Fibonacci {  
   public static int compute(int n) {  
     int result = 0;  
     if (n <= 1) {   
       result = n;   
     } else {   
       result = compute(n - 1) + compute(n - 2);   
     }  
     return result;  
   }  
 }  

The Golang way

 package Fibonacci  
 import "testing"  
 func TestFibonacci(t *testing.T) {  
      parameters := []struct {  
           input, expected int  
      }{  
           {0, 0}, {1, 1}, {2, 1}, {3, 2}, {4, 3}, {5, 5}, {6, 8},  
      }  
      for i := range parameters {  
           actual := Fibonacci(parameters[i].input)  
           if actual != parameters[i].expected {  
                t.Logf("expected%d: , actual:%d", parameters[i].expected, actual)  
                t.Fail()  
           }  
      }  
 }  
 func Fibonacci(n int) (result int) {
 if n <= 1 {
  result = n
 } else {
  result = Fibonacci(n-1) + Fibonacci(n-2)
 }
 return
}

MySql huge tables chunked delete

In order to avoid table locking while deleting data from huge tables, it's recommended to delete chunks of rows.


The following code also adds two additional gourds.

•     Date guard: It's impossible to delete data newer than 3 months

•     Time guard: If the process takes more than 10 minutes it quite.

1. First, we create a dynamic query utility procedure.

 CREATE PROCEDURE `dynamicQuery`(query_string text)

 BEGIN

     set @st=query_string;

     PREPARE stmt FROM @st;

     EXECUTE stmt;

     DEALLOCATE PREPARE stmt;

 END

2. For debugging we'll add a log table.

 CREATE TABLE `log` ( 

  `min` int(11) DEFAULT NULL, 

  `max` int(11) DEFAULT NULL, 

  `target` date DEFAULT NULL, 

  `scan` date DEFAULT NULL, 

  `scanid` int(11) DEFAULT NULL, 

  `counter` int(11) DEFAULT NULL, 

  `created` timestamp NOT NULL DEFAULT CURRENT_TIMESTAMP, 

  `procid` int(11) NOT NULL DEFAULT '0', 

  `object` varchar(63) DEFAULT NULL 

 ) ENGINE=MyISAM; 

3. Create delete_until procedure.

 CREATE PROCEDURE `delete_until`(tbl_name varchar(63),until_date DATE)  
 main: BEGIN  
   SET @TNAME=tbl_name;  
   SET @TARGET=until_date;  
   SET @MX=0;  
   SET @MN=0;  
   SET @BATCH=7000;  
   SET @VALIDTARGET=1;  
   SET @DELETED=-100;  
   SET @PROC=connection_id();  
   SET @ETSTART=UNIX_TIMESTAMP();  
   SET @TIMEOUT=600; -- 10 minutes  
   -- 3 month guard. Its impossible to delete data for the last three month  
      if datediff(date(now()),@TARGET) < 91 then  
           select 'only delete data that is older then three month';  
           LEAVE main;  
   end if;  
      SET @DELETED=0;  
      -- Get min and max id  
      SET @q=CONCAT("SELECT max(id), min(id) into @MX,@MN from ",@TNAME);  
      call dynamicQuery(@q);  
      -- Check the min id is older then @TARGET  
      SET @cond=CONCAT("SELECT EXISTS ( SELECT id FROM ",@TNAME," WHERE id = @MN AND created_at < @TARGET ) into @VALIDTARGET;");  
      call dynamicQuery(@cond);  
      -- Loop from min to max in 7000 batch  
      deleteloop: WHILE @VALIDTARGET>0 DO  
           -- Delete rows   
           SET @q=CONCAT("delete from ",@TNAME," where id between @MN and @MN+@BATCH and created_at < @TARGET;");  
           call dynamicQuery(@q);  
           -- Accumulate rows for summary  
           SET @DELETED = @DELETED+@BATCH;  
           -- Promote @MN Iterator  
           SET @MN=@MN+@BATCH+1;  
           -- If the largest id in the next iteration is newer then @TARGET, exit condition is met. @VALIDTARGET=0.  
           call dynamicQuery(@cond);  
           -- insert to log  
           insert into log(object,procid,min,max,target,scan,scanid,counter) values(@TNAME,@PROC,@MN,@Mx,@TARGET,null,@VALIDTARGET,@DELETED);  
           SET @ETEND=UNIX_TIMESTAMP();  
           if @ETEND-@ETSTART > @TIMEOUT then -- Ten minutes timeout guard exceeded.  
                LEAVE deleteloop;  
           end if;  
      END WHILE;  
      SELECT @DELETED;  
 END  

4. A chunk insert as a bonus.

 CREATE PROCEDURE `chunk_insert`(tbl_source varchar(63),tbl_target varchar(63)) 

 BEGIN 

   SET @TOTABLE=tbl_target; 

   SET @FROMTABLE=tbl_source; 

   SET @MX=0; 

   SET @MN=0; 

   SET @BATCH=7000; 

   SET @TOID=0; 

   SET @LASTLOOP=0; 

   SET @INSERTED=0; 

   SET @PROC=connection_id(); 

      -- Get min and max id from remote table 

   SET @q=CONCAT("SELECT max(id), min(id) into @MX,@MN from ",@FROMTABLE); 

   call dynamicQuery(@q); 

   -- Loop from min to max in 7000 batch 

   WHILE @LASTLOOP=0 DO 

           IF @MN+@BATCH > @MX THEN 

                SET @LASTLOOP=1; 

                SET @TOID=@MX; 

       SET @INSERTED=@INSERTED+(@MX-@MN); 

           ELSE 

                SET @TOID=@MN+@BATCH; 

       SET @INSERTED=@INSERTED+@BATCH; 

           END IF; 

     insert into log (procid,min,max,target,scan,scanid,counter) values(@PROC,@MN,@Mx,null,null,@TOID,@LASTLOOP); 

           SET @q=CONCAT('insert into ',@TOTABLE,' select * from ',@FROMTABLE,' where id between @MN and @TOID;'); 

     call dynamicQuery(@q); 

     SET @MN=@TOID+1; 

   END WHILE; 

   select @INSERTED; 

 END 

5. Usage: Delete data that is 5 months old from table `my_huge_table_name`

call delete_until('my_huge_table_name',DATE(NOW() - INTERVAL 5 MONTH));

MySql dump - exclude tables by pattern

In this example we want to ignore tables that prefix with lhm_ or _(underscore)

Create bash script with the following lines

#Generate --ignore-table clause
IGNORE=`mysql -A -N -umyuser -pmypass -hmyserver -e"select group_concat(DISTINCT d.ig ORDER BY d.ig ASC SEPARATOR ' ') from (select concat('--ignore_table=myDatabase.',table_name) as ig from information_schema.tables where (table_name like 'lhm_%' or table_name like '\_%')  and table_schema='mydatabase') d;"`


#Run mysqldump with the generated $IGNORE
mysqldump -umyuser -pmypass -hmyserver --routines --no-data --compact --triggers --hex-blob $IGNORE mydatabase -r 'dump.sql'

Hebrew encoding

Fix encoding

node-iconv -t UTF-8 -f ISO-8859-8 in.srt > out.srt

Here are some guidelines to choose the from (-f) encoding 

בשלוש הדוגמאות האלו רואים עברית הפוכה. שימו לב שבדוגמה הראשונה סימן הקריאה הוא בצד ימין ובשתי הדוגמאות האחרונות הוא בצד שמאל. 

 זה קידוד CP-1255 הוצג כ- ISO-8859-8 

 זה ISO-8859-8 הוצג כ- CP-1255 

 זה ISO-8859-8 הוצג כ- MacHebrew 

בעית קידוד נוספת שגורמת לתופעה דומה היא ציון ISO-8859-8 במקום ISO-8859-8-I או להפך. אז תבדקו גם את האפשרות הזאת.

ארבע הדוגמות הבאות כוללות ביתים שגויים, לכן הופעת סימן היהלום. שימו לב שרק בדוגמה האחרונה מופיע סימן הקריאה בצד שמאל. 

 זה UTF-8 הוצג כ- CP-1255 

 זה UTF-8 הוצג כ- ISO-8859-8 

 זה CP-1255 הוצג כ- UTF-8 

 זה ISO-8859-8 הוצג כ- UTF-8

המצב הזה קל לזיהוי בגלל הצלבים המפרידים בין סימן לסימן. 

 זה UTF-8 הוצג כ- ISO-8859-1

הכתב ה"שבדי" הוא תוצאה של הצגת קידודים אחרים בתור קידוד לטיני: 

 זה CP-1255 הוצג כ- ISO-8859-1 

 זה ISO-8859-8 הוצג כ- ISO-8859-1

מצב זה הוא די נדיר. כאן רואים קידודים שונים המוצגים כקידוד של י.ב.מ.

 זה CP-1255 הוצג כ- IBM-862 

 זה ISO-8859-8 הוצג כ- IBM-862

שני מצבים אלו גם כן נדירים. שניהם נגרמים כתוצאה מהצגת UTF-8 בתור קידודים שונים. 

 זה UTF-8 הוצג כ- IBM-862 

 זה UTF-8 הוצג כ- MacHebrew

http://gibberish.co.il/encoding.html

Make current directory DocumentRoot in apache

Create the following bash file and put it in the directory you want to use as apache root directory

#!/bin/sh
sed 's!\DocumentRoot .*!DocumentRoot '"`pwd`"'!' /etc/apache2/sites-available/default > default
sudo mv default  /etc/apache2/sites-available/
sudo /etc/init.d/apache2 restart

Connect Oracle with CakePHP

Tested with Windows XP and XAMPP 
1. Install XAMPP
2. Install OracleXEUniv
3. In php.iniuncomment the following line:
     extension=php_oci8.dll
4. Set app/config/database.php :
var $default = array('driver' => 'oracle',
             'connect' => 'oci',
            'persistent' => false,
            'host' => 'localhost',
            'port'=>1521,
            'login' => 'root',
            'password' => 'pass',
            'schema'=>'schama_name',
            'database' => 'localhost:1521/xe',
            'prefix' => ''
         );

Connect Cakephp application to remote database via SSH

Tested on ubuntu 10.4
Edit /etc/ssh/sshd_config on your local machine
Add:
AllowTcpForwarding yes
GatewayPorts yes

restart sshd
sudo /etc/init.d/ssh restart


Edit /etc/mysql/my.cnf on remote machine.
Uncomment the following line.
bind-address           = 127.0.0.1

restart mysql
sudo restart mysql

On you local machine run the following command
ssh -i e-keypair -f remoteuser@remote.com -N -L 4444:localhost:3306
If you are not using ssh key you will be prompt for remote user assword

Edit database.php

var $default = array(
        'driver' => 'mysqli',
        'persistent' => false,
        'host' => '127.0.0.1',
        'port'=>4444,
        'login' => 'user',
        'password' => 'pass',
        'database' => 'schema',
        'prefix' => '',
        );